Environmental and Experimental Botany 176 (2020) 104085

Contents lists available at ScienceDirect

Environmental and Experimental Botany

journal homepage: www.elsevier.com/locate/envexpbot

Potassium fertilization increases hydraulic redistribution and water use eﬃciency for stemwood production in Eucalyptus grandis plantations T

Verónica Asensioa,*, Jean-Christophe Domecb,c, Yann Nouvellond,e, Jean-Paul Laclaud,e, Jean-Pierre Bouilletd,e, Lionel Jordan-Meilleb, José Lavresa, Juan Delgado Rojasf, Joannès Guillemotd,e,g, Cassio H. Abreu-Juniora a Universidade de São Paulo-Centro de Energia Nuclear na Agricultura (USP-CENA), CEP 13400-970 Piracicaba, SP, Brazil b Bordeaux Sciences Agro, INRA UMR 1391 ISPA, F-33170 Gradignan, France c Nicholas School of the Environment, Duke University, Box 90328, Durham, NC 27708, USA d Eco&Sols, Univ Montpellier, CIRAD, INRA, IRD, Montpellier SupAgro, 34060 Montpellier, France e CIRAD, UMR Eco&Sols, 34060 Montpellier, France f Agro Ambiência Serviços Agrícolas, Piracicaba, SP, Brazil g Department of Forest Sciences, Universidade de São Paulo-“Luiz de Queiroz” College of Agriculture (USP-ESALQ), CEP 13418-900 Piracicaba, SP, Brazil

ARTICLE INFO ABSTRACT

Keywords: Climate change is expected to increase the frequency of droughts in most tropical regions in the coming decades. Eucalyptus A passive phenomenon called hydraulic redistribution (HR) allows some plant species to take up water from Hydraulic lift deep and wet soil layers and redistribute it in the upper dry layers where other plants and soil biota can benefit Soil fertilization from it. In addition, soil fertilization, particularly potassium (K), may also affect drought-adaptive mechanisms Tree transpiration and increase water use efficiency (WUE) on poor and acidic tropical soils. The present study aimed at quantifying Water use the role of HR and K fertilization on both wood productivity and WUE for stemwood production (WUEp)of Eucalyptus grandis plantations in Brazil under ambient and reduced (−37%) throughfall conditions. Tree transpiration was measured using trunk sap flow sensors over 21 months, and HR was estimated from the reverse sap flow (RF) observed in shallow roots over 18 months. Tree biomass, hydraulic conductance, soil water storage from surface to the water table (down to 17 m), and leaf photosynthetic capacity were also assessed. Significant HR was detected over the whole year, even during the rainy seasons. Neither potassium fertilization nor throughfall exclusion affected the velocity of water transported by HR, probably because most trees reached water table. Nonetheless, some photosynthetic capacity parameters, including the maximum photo-

synthetic rate (Amax), increased in treatments with K addition. This higher Amax combined with an increased sapwood area index, was associated with an increase in water uptake by 30 %–50 % and WUEp by 300 % relative

to K-deﬁcient trees. We postulate that the increase in WUEp promoted by potassium fertilization was partly driven by an increase in biomass allocation to wood, at the expense of foraging organs (leaves and roots), because K addition alleviated constraints on light and water use. Our results indicate that fertilizing E. grandis

plantations with K is beneﬁcial to both wood biomass production and WUEp.

1. Introduction and water vapor pressure deficit, and an increase in the length of dry periods (Chou et al., 2012; Salazar et al., 2016). From 2012–2015, the Extreme climatic events such as heavy rainfall, wind storms, and state of São Paulo, in southeast Brazil, suffered a 20–23 % reduction in severe droughts are increasing in frequency worldwide as a con- precipitation, which led to a significant decrease in soil moisture sequence of climate change (IPCC, 2014; Sherwood and Fu, 2014). (Getirana, 2016). The increase in the frequency of extreme climatic Recent climate predictions show that Southeast Brazil will be affected events will increase the difficulty for plants to acclimate physiologically by more frequent heavy rainfall events, higher mean air temperatures and morphologically to the rapidly changing climatic conditions, which

⁎ Corresponding author. Present address: Department of Soil Science, Universidade de São Paulo-“Luiz de QueirozCollege of Agriculture (USP/ESALQ), CEP 13418- 900, Piracicaba, SP, Brazil. E-mail address: [email protected] (V. Asensio). https://doi.org/10.1016/j.envexpbot.2020.104085 Received 30 January 2020; Received in revised form 15 April 2020; Accepted 15 April 2020 Available online 16 May 2020 0098-8472/ © 2020 Elsevier B.V. All rights reserved. V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085 will likely result in decreased primary productivity and increased (Stape et al., 2004, 2010; Binkley et al., 2017). Their fast growth is mortality (Choat et al., 2018). Since planted forests play an important often associated with high water-use, thus raising concerns on their role in supplying human needs for wood and paper products, it is im- potential impact on water resources. There is therefore much interest in portant to determine which forest management practices would be assessing their water-use eﬃciency (WUE), a highly integrative in- more appropriate for trees to grow properly under those new climatic dicator related to the plantation ability to capture carbon and produce conditions (Noormets and Nouvellon, 2015). biomass per unit of water use (Farquhar et al., 1989; Law et al., 2002; High mortality rates in commercial eucalypt plantations during re- Mateus et al., 2019). At the stand level, WUE for stemwood production cent droughts in Brazil (Scolforo et al., 2019a) have highlighted the (WUEp) can be computed as the ratio of stemwood biomass increment need to understand the key physiological processes that allow trees to to the water transpired over the same period. In fact, in areas with acclimate to recurrent dry summers (Lobell et al., 2011; Gessler et al., limited water resources, managing planted forests in a way that in-

2017). A suitable nutrient application, and in particular potassium (K) creases tree WUEp is a promising avenue to sustain productivity under fertilization has been shown to have positive effects on plant water and climate change. light use efficiency during drought periods (Harvey and van den The main goal of our study was to assess the effects of K fertilization Driessche, 1997; Cakmak, 2005; Battie-Laclau et al., 2016). Previous on wood production, water use, hydraulic redistribution between soil studies have evaluated the effect of the interaction between K fertili- layers, whole tree hydraulic conductance (Ktree) and WUEp in Eucalyptus zation and water availability on plant growth, but most of them with grandis plantations grown under ambient and reduced rainfall condi- non-woody species such as corn and sugarcane (da Silva Moura et al., tions, over the three last years of a commercial rotation (from 4 to 6 2005; Djaman et al., 2013; Mahmood et al., 1999, Martineau et al., years after planting). We hypothesized that (i) HR occurs in the studied 2017). Other studies on woody species like olive trees, were performed Eucalyptus grandis plantations, and is more pronounced under reduced on small and potted individuals growing under greenhouse conditions soil water availability; (ii) K addition increases HR and; (iii) K addition (Arquero et al., 2006). Under field conditions, the only studies ad- increases water-use efficiency for stemwood production. While Battie- dressing the potassium nutrition x drought interacting effects on woody Laclau et al. (2016) studied the effects of K fertilization and throughfall species were conducted on Eucalyptus grandis in Brazil (Battie-Laclau exclusion on WUEp over the early growth of the trees in the same ex- et al., 2014; Christina et al., 2015) but were limited to the first three periment, we show in this paper stand water use and WUEp at the end of growing seasons after planting. The end of the rotation cycle (that ty- the rotation cycle, as well as HR and whole-tree water transport (tree pically lasts 6–7 years) is however a crucial period, when most of the hydraulic conductance; Ktree), which were not evaluated in previous drought-related mortality in Brazilian eucalypt plantations occurs studies. (Scolforo et al., 2019b). This is likely due to decreasing soil water availability and increasing water table depth as stand ages (Christina 2. Materials and methods et al., 2018). Potassium fertilization was shown to increase photosynthesis and tree biomass during the first years after harvest (Battie- 2.1. Experimental design Laclau et al., 2014), but it also increased tree transpiration, thus leading to faster soil water depletion and increased plant water stress during the The experiment was carried out at the Itatinga Experimental Station dry seasons (Battie-Laclau et al., 2014). An evaluation of the interaction of the University of São Paulo in Brazil (23°02´S; 48°38´W). Over the effect of K fertilization and soil water availability on the water use and previous 15 years, the mean annual rainfall was 1360 mm. The mean water-use efficiency of eucalyptus trees along a whole commercial ro- temperature was 15 °C during the dry season (from June to September), tation is thus strongly needed. and 25 °C during the rainy season (from October to May). The experi- Plants have different strategies for surviving during drought per- mental site was located on a hilltop (slope < 3 %) at an altitude of 850 iods. Some plant species such as Eucalyptus trees can take up water from m. The soils were very deep (> 15 m) Ferralsols, with a clay content deep and moist soil layers through deep roots and then transport it up ranging from 14% in the A1 horizon to 23 % in the deep soil layers. The to dry soil layers, where it can be used when transpiration demand mineral content was dominated by quartz, kaolinite and oxyhydroxides increases (Burgess et al., 1998, 2001; Burgess, 2006; Brooksbank et al., and the soil was acid (pH 4.5–5). Before the experiment, exchangeable −1 2011). This process is called hydraulic redistribution (HR), or specifi- K and Na concentrations were on average 0.02 cmolc kg in the upper -1 cally hydraulic lift (HL, Weaver, 1919; Richards and Caldwell, 1987) soil layer and < 0.01cmolc kg from 5 cm to 1500 cm (Laclau et al. when water is only moved upward in response to water potential gra- 2010). The rhizosphere biogeochemistry was studied down to a depth dients (ΔΨ) in the soil-plant system, occurring when Ψsoil in shallow of 4 m in this experiment (Pradier et al., 2017), as well the vertical soil layers is more negative than Ψsoil in deep soil layers and when distribution of ectomycorrhizas (Robin et al., 2019). Meteorological stomata are closed (Domec et al. 2004; Scholz et al., 2008; David et al., data were measured during the whole experiment using an automatic 2013). Both HR and HL usually occur at night, when leaf transpiration weather station located at the top of a 21 m tower, at 50 m from the has diminished, and during dry seasons, when soil water content is low experimental plots. in upper soil layers. This process was shown to improve plant growth A split-plot experimental design, described in detail in Battie-Laclau and survival during droughts (Brooks et al., 2002; Scholz et al., 2008; et al. (2014), was set up in June 2010 with a highly productive E. Bleby et al., 2010). However, few studies directly measured HR and HL grandis clone used in commercial plantations by the Suzano Company on roots with sap flow sensors (Scholz et al., 2002; Brooks et al., 2002; (Brazil). Two K fertilization regimes (+/-K) and two water supply re- Domec et al., 2010; Bleby et al., 2010; Nadezhdina et al., 2015). All gimes (+/- W) were applied in three blocks, thus leading to four these studies except for Bleby et al. (2010) and Nadezhdina et al. (2015) treatments and twelve plots. The four treatments were: monitored root sap flow only during the dry season. Bleby et al. (2010) -K + W: without K fertilization, and no throughfall exclusion, − observed reverse flow in tree roots of a semi‐arid woodland during the +K + W: control, 0.45 mol K m 2 applied as KCl and no whole year. Monitoring HR over the whole growing season appears throughfall exclusion. therefore important to understand the drivers of tree water use and -K-W: without K fertilization, and ∼37% of throughfall excluded, − acclimation to drought. +K-W: 0.45 mol K m 2 applied as KCl, non-limiting in terms of the Subtropical and tropical hardwood plantations are dominated by availability of K for tree growth (Almeida et al., 2010), and ∼37% of the Eucalyptus genus. Within this genus, the highly productive E. grandis throughfall excluded, is the most planted species worldwide in moist and warm subtropical The area of each individual plot was 864 m2, with 144 trees per plot regions (Gonçalves et al., 2013). A strong effect of rainfall regimes on at a spacing of 2 m x 3 m (1667 trees per hectare), except one plot for the productivity of eucalypt plantations has been widely documented each treatment that was 50% larger to allow destructive sampling

2 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085 without disturbing tree growth in the inner plots where measurements were installed at 1.3 m above the ground and protected from external were made. The KCl fertilization was applied once, 3 months after temperature variations and water intrusion by a reflective aluminum planting. All the plots in the experiment were fertilized with other foil. The sensor output voltage was recorded every 30 s and the 30-min − nutrients when the trees were planted (3.3 g m 2, 200 g m-2of do- average values stored in data loggers (CR1000, Campbell Scientific Inc., − lomitic lime, and micronutrients) and after 3 months (12 g m 2). Field Logan, UT, USA). All sensors were checked every week and replaced if trials at the study site and in nearby areas on the same type of soil they were physically damaged. Additionally, the sensors were moved to showed that the amounts of N, P, Ca, Mg and micronutrients applied in newly drilled holes in another position round the trunk (selected at this experiment were not limiting for tree growth (Gonçalves et al., random) 3 months after installation to prevent sap flow from being 2008; Laclau et al., 2009). underestimated due to fast radial growth. The CBH of all equipped trees Throughfall was partially excluded in the -W plots from September was measured monthly. The sap-flow sensor signal was converted to sap −2 −1 2010 onwards, using panels made of clear greenhouse plastic sheets flux density (SFD, kg_H2Om _sapwood s ) using the calibration mounted on wooden frames at heights varying between 0.5 and 1.6 m. equation determined previously at our site on 18 trees (Delgado-Rojas 2 Plastic sheets covered 37% of the area in the -W plots leading to a et al., 2010) and the sapwood area at breast height (As, m ). The semi- −1 −1 throughfall exclusion of about ∼450 mm y in the -W plots. Scaffold hourly sap flux (kg_H2O 30 min ) was then calculated for each tree towers were used to access the crown of four trees in the central part of and then cumulated over the day to compute daily transpiration −1 each plot at block 1 (four towers in total). (kg_H20d ). Then, for each tree fitted with a sap flow sensor in each treatment, the relationship between the daily total transpiration (de- 2 2.2. Tree biomass and leaf area index pendent variable) and CBH (independent variable) was determined by linear regression (SFD = a+b·CBH2), as in other recent studies (Kunert 2 Tree height and circumference at breast height (H and CBH, re- et al., 2012; Battie-Laclau et al., 2016). The mean R values of all the 2 spectively) were measured every 6 months on 36 trees per plot (ex- regressions between daily total transpiration and CBH over the study cluding a minimum of three buffer rows to avoid edge effects), except period were between 0.60 and 0.71 in the four treatments. These re- for the last year, when only one inventory was made at the end of the gressions were then used to estimate the daily stand transpiration (E, -1 rotation. The aboveground biomass was estimated at 47, 59 and 71 mm day ) from the CBH of all the trees in each inner subplot (linearly months by sampling eight trees representative of the range of cross- interpolated between 2 measurements of CBH per year), which made it sectional areas in each treatment. The selected trees were felled and possible to take into account the changes in distribution of CBH in each ffi measured for total height (Ht) and height of the crown base (Hb). Then stand over the study period. Then the water use e ciency for stemwood the crown length (L = Ht -Hb) was divided into three equal-length production (WUEp) was calculated in each inner subplot as the stem- sections (lower, middle and upper). The foliage biomass was de- wood biomass produced divided by the total amount of water tran- termined by weighing all the leaves from each of the three crown spired over the same period. Since tree biomass was measured at times sections, and randomly subsampling 30 leaves per crown section. that do not match the beginning or the end of sap flow measurements, Sampled leaves were immediately scanned at 300 dpi and fresh mass stand stemwood biomass for August 2014, April 2015 and April 2016 was measured. Leaves were then oven dried at 65 °C for three days, and was calculated using CBH values interpolated between two consecutive weighed again. The dry weights of the sub-samples were used in con- inventories, based on the monthly CBH records available for the trees junction with their measured area to calculate specific leaf area (SLA) equipped with sap-flow sensors. for each crown section. The foliage dry weight of each crown section The sapwood area at 1.3 m above the ground was determined by was calculated from the foliage fresh weight and the dry to fresh weight pressurizing water dyed with concentrated liquid dye into the trunk ratio of the sub-samples. The tree leaf biomass (Bl) and leaf area (Al) section of 8 trees felled at 4, 5 and 6 years in each treatment. The al- were then estimated as: lometric relationships between tree CBH and sapwood area were determined for each treatment and each age and applied to the CBH in- 3 BB= ventories of all trees of the experiment to calculate sapwood area index l ∑ i 2 −2 fl i=1 (cm _sapwood m _ground). Sap ow of each tree (SF) was calculated as the product of SFD and sapwood area estimated from successive 3 measurements of CBH over the study period. The sensors were assumed ABSLA= l ∑ ii to measure the instantaneous sap velocity integrated over the average i=1 sapwood thickness of each tree. It was not necessary to apply a cor- 2 −1 where Bi and SLAi are the foliage dry weight (kg) and SLA (m kg )in recting factor to extrapolate sap flow density to whole-tree sap flux the ith crown section, respectively. (Delzon et al., 2004) because the sapwood areas of the measured trees Live branches, dead branches, stemwood (diameter > 2 cm at the never exceeded the length of the sap flow sensor (2 cm). thinner end) and stem bark were collected in the field for all the trees and weighed separately. Subsamples were taken from all the tree 2.4. Root sap flow compartments (live branches, dead branches, stemwood and bark), dried at 60 °C until constant weight, and the dry biomass of the com- To detect the relative influence of transpiration and HR on root sap ponents in each tree was calculated proportionally. flow, root SF was monitored from November 2014 to April 2016 (18 Treatment-specific allometric relationships were established at each months) in six roots per treatment, corresponding to one root in each of age and applied to the inventory made on the same date to estimate the trees selected for monitoring sap flow. The excavated roots were −2 plot-scale above-ground biomass (kg m ) and leaf area index (LAI, shallow (< 20 cm deep), horizontal and thicker than 1-cm in diameter 2 −2 m _leaf m _ground) from tree attributes (diameter at breast height and sampled at least 2 m away from the tree trunk to make sure that and height); see Battie-Laclau et al. (2016) for further details on the they were located in the upper soil. The space of the root with installed allometric equations and the fitting procedures. sensors was protected from sunlight and rainfall with a plastic cap covered with a reflective foil. 2.3. Stand transpiration and water use efficiency for biomass production Root SF was measured with the same thermal dissipation technique used to determine tree transpiration but modified to allow the direction Trunk sap flow was continuously measured between August 2014 of flow to be detected (Brooks et al., 2002; Domec et al., 2010). A di- and April 2016 (for 21 months) in 6 trees per treatment using the heat rectional probe was installed in each selected root to detect reversal of dissipation technique (Granier, 1985). Two sap flow sensors per tree flow. The heated temperature sensor was 10 mm in length, and it was

3 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085 inserted into the xylem in the center of the exposed root. Two unheated There was a loss of data from February to October 2015 for treatments reference temperature sensors, also 10 mm in length, were placed -K + W and -K-W due to technical problems. axially 8 cm up- and downstream from the heater sensor and wired to The depth of the water table was monitored at the subplots of + K measure the temperature differences between the heated and unheated + W and + K-W in block 1 with piezometers located 18.9 and 19.9 m sensors. For the directional probe, two thermocouples were inserted 5 deep, respectively. Data were collected every hour and showed that the mm axially to a depth of 5 mm up- and downstream from the heated water table varied from 18.01 to 16.52 m in + K + W and from 17.06 probe (Brooks et al., 2002). The movement of the heated water in- to 17.16 m in + K-W, from the beginning to the end of the study. creases the temperature of the thermocouple closest to the trunk relative to that of the distal thermocouple, providing an accurate measure 2.7. Photosynthetic measurements to direction of flow. Sap flow moving toward the tree was indicated as positive, whereas sap flow moving away from the tree (reverse flow) Photosynthetic net CO2 assimilation rate as a function of inter- was indicated as negative. The temperature of the reference probe that cellular CO2 (Ci) concentration (A-Ci curves; Long and Bernacchi, was determined to be upstream of the direction of flow was used to 2003)) was measured on 1- to 2-month-old leaves. Measurements were calculate root SF when the direction of the flow was positive, and the carried out in four leaves per treatment with a Li-Cor 6400 portable downstream reference probe when there was reverse flow. Root sap- photosynthesis system (Li-Cor Inc., Lincoln, NE, USA) on May 2016. wood area was determined by injecting dye with water pressure into During field measurements, the CO2 concentration in the gas exchange the section with sensors of each monitored root. The whole section was chamber was reduced from 400 to 300, 250, 200, 150, 100, 75 and 50 −1 stained in all roots, indicating that the whole area was sap-conducting. μmol CO2 mol , and then increased from 50 to 400, 600, 800, 1000, −1 The reverse volumetric flow rate was calculated as the average of RF 1300 and 1600 μmol CO2 mol . These measurements were carried out velocity (i.e., averaging all RF velocity > 0) over each month. at a constant photosynthetic photon flux density (PPFD) of 1600 μmol − − − − The sensor signal was converted to SF (cm3 cm 2 30 min 1) using m 2 s 1, ambient relative humidity (60–75 %) and leaf temperature of the calibration equation determined for stems (Delgado-Rojas et al., 25 °C. From these A-Ci curves, the maximum photosynthetic rate at −1 2010), without taking into account sapwood area. These semi-hourly ambient CO2 concentration (400 μmol CO2 mol ) and saturating PPFD values were summed up over 24 h to obtain a daily value, accounting (Amax), the mesophyll conductance to CO2 (gm), the maximum rate of positive values as regular flow and negative values as reverse flow. RubiscO-catalysed carboxylation (Vc,max), the ribulose bisphosphate (RuBP) regeneration rate controlled by the electron transport rate at

2.5. Leaf and whole tree hydraulic conductance saturating PPFD (Jsat), and the RuBP regeneration rate controlled by the triose phosphate utilization (TPU), as defined by Farquhar et al. (1980) Leaf water potential was measured monthly from October 2014 to and Harley et al. (1992), were estimated using the Sharkey et al. (2007) May 2016 (20 months) in each treatment on four of the same trees used curve fitting utility software (version 1.2). for sap flow (total of 16 sampled trees) with a Scholander-type pressure chamber (PMS Instrument Company, Albany, OR, USA). Measurements 2.8. Statistical analyses were carried out on one fully expanded leaf (approximately two months old and accessed through a scaffold tower) sampled in the upper third In order to explore the significance of the treatment effects on the of the crown of each tree. Predawn leaf water potential (Ψpdwn) was studied variables, mixed-effect models were used. The water supply measured before sunrise between 6h00 and 7h00. On the same day and (W), fertilization (F) treatments, as well as stand age, season (wet vs on the same four trees, midday leaf water potential (Ψmd) and branch dry), W x F, W x stand age, F x stand age and W x K x stand age were water potential (Ψbranch) were measured between 12h00 and 14h00. included as fixed effects. Studied variables were stand transpiration, Ψbranch was determined from leaves covered with a sealed plastic bag WUEp, LAI, stand transpiration per m²leaf, wood increment per LAI, and aluminum foil installed the evening before the measurements. leaf area per sapwood area ratio, leaf-specific Ktree, sapwood, sapwood- Under these conditions, leaves are not transpiring and the leaf water specific Ktree, daily reverse flow velocity, daily reverse volumetric flow potential is generally agreed to equilibrate to that of the adjacent xylem rate, number of days with reverse flow, and number of roots with RF. (Melcher et al., 1998). Blocks and water supply x block were considered as random effects,

Field leaf hydraulic conductance (Kleaf) was calculated as Kleaf = E/ consistently with the split-plot design of the experiment, and the re- (Ψbranch – Ψmidday), where E is the value of transpiration measured with siduals were modeled by a first-order autoregressive correlation model sap flow sensors converted to a tree-scale average transpiration per unit to account for the correlations between sampling dates. For reverse flow − − leaf area (E, in mmol m 2 s 1)(Domec et al., 2009). We used the variables, monitored roots and water supply x roots were considered as transpiration values from the same day as Ψmidday and Ψbranch were random effects. In order to evaluate weather treatment modalities had measured, from 12:30 h to 14:00 h. The leaf-specific tree hydraulic significantly different values for a given variable, post-hoc tests were conductance (leaf-specific Ktree) that represents the leaf-specific hy- done. In case of homogeneity of variance, a least significant difference draulic conductance of the soil to tree to atmosphere continuum was test (LSD) was carried out. If there was no homogeneity, Dunnett's T3 calculated according to Ohm’s law (Cochard et al., 1996, Loustau et al. test was performed. A value of 0.05 was used as the level of significance

1998) as Ktree = E/(Ψpdwn – Ψmidday). The sapwood-speciﬁc Ktree was for all tests. All statistical analyses were carried out using the software calculated with the same equation than for leaf-speciﬁc Ktree but nor- SPSS Statistics for Windows, version 15.0 (SPSS Inc., Chicago, Ill., malizing E for sapwood area index instead of leaf area index. USA).

2.6. Soil water storage and water table 3. Results

Soil volumetric water content was measured with three TDR probes 3.1. Tree transpiration per soil layer (CS616 and CS650, Campbell Scientific Inc., Logan, UT, USA) installed at depths of 0.15, 0.50, 1, 1.5, 3, 4.5, 6, and 8 m in each While K fertilization greatly increased tree transpiration, the effect subplot in block 1. In addition, 3 probes CS616 were also installed at of throughfall reduction was not significant (Fig. 1, Table 1). The 10, 12, 14 and 17 m depth in the subplots of the treatments + K + W amounts of water transpired by Eucalyptus grandis trees were 30 % to 50 and + K-W. Probes were calibrated by gravimetric soil water content % higher in the plots fertilized with K (+K) than in the unfertilized and bulk density measurements. Volumetric soil water contents were plots (−K) under ambient precipitation regime (water control treat- continuously measured every 30 s from August 2014 to April 2016. ment; +W) over the studied period (Fig. 1a, Fig. S1, Table S1). Under

4 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085

always lower than in the +K + W plots, and ii) this difference increased over time (Fig. 1b). Stand transpiration was significantly influenced by stand age (Table 1). The ratio of stand transpiration between +K−W and the control treatment (+K + W) shows that throughfall exclusion decreased the transpiration in plots fertilized with K, especially over dry periods. Stand transpiration was significantly influenced by the seasons (Table 1). At the end of the dry seasons or at the beginning of the wet ones, the ratios of stand transpiration between +K−W and −K treatments were similar (Fig. 1b). It was expected that the amount of water transpired be considerably higher in wet than in dry seasons, but it only happened in + K-W plots. In the other treatments, there were significant differences in transpiration only in some months (P < 0.05) (Table S1). It was observed that in + K + W, transpiration from month 62 to the end of the rotation was significantly higher than in previous months, even in the months of dry season (62 and 63). In all treatments, we detected small amount of night-time trunk water movement over the whole study period (Fig. S2).

3.2. Water use eﬃciency of Eucalyptus

Tree transpiration was positively correlated to wood increment for + K treatments only (P < 0.05) (Fig. 2a). K deﬁciency strongly de-

creased WUE for wood production (WUEp): in -K plots WUEp was only a quarter of that in + K plots (Fig. 2b). By contrast, WUEp was not significantly affected by throughfall exclusion whatever the K supply regime (Fig. 2b, Table 1). Potassium significantly increased leaf area (LAI) and sapwood area Fig. 1. (a) Monthly rainfall and stand transpiration in Eucalyptus grandis plots at index (Table 1) (Fig. S4). Trees in + K plots exhibited higher leaf area fl the Itatinga experimental site, and (b) ratio between the monthly sap ow to sapwood area ratios than trees in -K (Fig. 3a). On the other hand, measured in treatment plots (- K + W, -K-W, +K-W) and in control plots (+K those trees fertilized with K showed lower tree water use per unit of leaf + W), from August 2014 to April 2016 (50–70 months after planting). Each area (Fig. 3b), and higher growth efficiency (wood production per leaf point in figure “a” is the sum of 30–31 daily values (SE for monthly transpiration were always lower than 0.50 mm and are not shown on the figures). area; Fig. 3c), than trees in -K plots. Neither potassium fertilization nor fi ff Applied treatments were: K fertilization and undisturbed rainfall as in adjacent throughfall exclusion had signi cant e ect on leaf or whole-tree hy- – commercial plantations (+K + W, control), neither K fertilization nor draulic conductances (Kleaf and Ktree, respectively; in Fig. 4a c). throughfall exclusion (-K + W), no K application and 37% throughfall exclusion (-K-W), K fertilization and 37% throughfall exclusion (+K-W). 3.3. Root sap flow and hydraulic redistribution (HR) over 18 months

37% throughfall exclusion (−W), the amounts of water transpired were Reverse sap ﬂow in roots (RF, away from the trunk) indicated the similar in plots with and without K addition (Fig. 1a, Table S1). Most of occurrence of HR. We detected RF every month for 18 continuous the time, tree transpiration was higher in +K−W than in −K+W, months in the + W plots (Fig. 5, Table S2). Roots in the -W plots although throughfall was reduced by 37 %. The ratios of monthly showed RF all months except in November and December of both years transpiration between plots without K fertilization (−K) and the con- in -K-W, and December 2015 and February 2016 in + K-W. Those trol treatment representative of nearby commercial plantations (+K + months corresponded to wet periods. We also observed that all mon- W) highlighted that: i) transpiration in plots without K fertilization was itored roots (6 per treatment) had RF, at least for a few days. In some days with high transpiration rates, we also detected RF during the day

Table 1 P-values for the effects of fertilization regime (F, control, supply of K), water supply regime (W, undisturbed rainfall vs 37% of throughfall exclusion), stand age (age), season (wet vs dry), interaction between water supply and fertilization (F x W), fertilization and stand age (F x age), water supply and stand age (W x age), among water supply, fertilization and stand age (W x F x age) on: stand transpiration, water use efficiency for stem wood production (WUEp), leaf area index (LAI), sapwood area index, leaf area per sapwood area (Al/As), sap flow per leaf area (SF/Al), wood increment per leaf area (wood/Al), daily reverse flow velocity (RF vel), daily reverse volumetric flow rate (RF vol), and number of days with reverse flow (RF days) for Eucalyptus grandis trees in a split-plot design.

F W F x W Age F x Age W x Age F x W x Age Season

Transpiration < 0.001 0.695 0.042 < 0.001 < 0.001 0.880 0.011 < 0.001

WUEp < 0.001 0.208 0.149 < 0.001 < 0.001 0.129 0.148 – LAI < 0.001 < 0.001 < 0.001 0.456 0.491 < 0.001 < 0.001 – Sapwood 0.007 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 – Al/As < 0.001 < 0.001 < 0.001 < 0.001 0.006 < 0.001 < 0.001 – SF/Al < 0.001 < 0.001 < 0.001 0.010 < 0.001 < 0.001 < 0.001 – Wood/Al 0.101 0.057 < 0.001 < 0.001 0.011 0.070 0.001 – RF vel 0.750 0.579 0.844 0.072 0.597 0.634 0.846 0.347 RF vol 0.367 0.758 0.339 0.034 0.418 0.817 0.383 0.126 RF days 0.354 0.141 0.204 0.757 0.257 0.208 0.222 0.002

Transpiration is shown in Fig. 1 and S1, WUEp in Fig. 2b, LAI and sapwood area in Fig. S4, leaf area per sapwood in Fig. 3a, sap flow per leaf area in Fig. 3b, wood increment per leaf area in Fig. 3c, daily reverse flow velocity in Fig. 5, daily reverse volumetric flow rate in Table S2, and number of days with reverse flow in Table S3. Dash lines means there is no factor “season” for those variables.

5 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085

Fig. 2. (a) Mean daily wood increment in relation to mean daily tree transpiration in each plot (3 plots per treatment, and 2 measurement periods), and

(b) water use efficiency for stem wood production (WUEp) of each treatment for the last two years before harvesting (5th and 6th). Calculations for the 5th year were made without the first three months of the year due to the unavailability of sap flow data. Values are the mean ± SE (n = 3). Within each year, bars with different letters are significantly different (P < 0.05). The treatments were: K fertilization and undisturbed rainfall (+K + W), no K fertilization and undisturbed rainfall (-K + W), no K application and 37% throughfall exclusion (-K-W), K fertilization and 37% throughfall exclusion (+K-W). in some roots, with sometimes maximum reverse flow occurring even at midday (Fig. S3). Neither potassium fertilization nor throughfall exclusion significantly affected RF velocity (Table 1). Nevertheless, we observed that – during the dry season (months 60 63), and month 70 (low rainfall), Fig. 3. Water and potassium fertilization effects on (a) leaf area index (LAI) per when is expected the highest RF of the year, daily RF was always higher sapwood area, (b) daily tree water use, and (c) daily wood increment on a leaf in + K plots than in -K plots, and it was generally higher in + W plots area basis 5 and 6 years after planting. Calculations for the 5th year were made than in -W plots (Fig. 5). Such increase in RF velocity during dry periods without the first three months of the year due to the unavailability of sap flow was especially evident in + K + W. data. Values are the mean ± SE (n = 3). Within each year, bars with different In order to observe the capacity of roots to transport water away letters are significantly different (P < 0.05). Applied treatments were: K ferti- from the trunk, their reverse volumetric flow rates were calculated lization and undisturbed rainfall (+K + W), no K fertilization and undisturbed rainfall (- K + W), no K application and 37% rainfall exclusion (-K-W), K fer- (Table S2). Neither potassium fertilization nor throughfall exclusion tilization and 37% rainfall exclusion (+K-W). had an effect on the reverse volumetric flow rate (Table 1). Never- theless, we observed that potassium fertilization tended to increase the reverse volumetric flow rate, especially in dry season (Table S2). The number of days with RF (Table S3) was also not influenced by

6 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085

season.

3.4. Soil water and water table dynamics

Soil water storage was higher in the -K plots than in the + K plots, from soil surface to 17 m in depth over the studied period (Fig. 6 a–d). Soil water storage was also higher under + W than under -W. It can be observed the diﬀerences in soil water content between treatments increased with soil depth. It can also be observed that from 7 to 17 m in depth, the soil water storage varied over time in -K plots, whereas it was more stable in + K plots. Water table level was always deeper in the throughfall exclusion plots than in the control plots (Fig. 6e). The deepest level of the water table during the experiment was measured at the end of the dry season (November of 2015) and reached 18 m in + K + W plots and 19 m in +K–W plots. The shallowest level was 16.5 m in + K + W and 17.2 m in + K–W, at the end of the rainy season 2016 (April).

3.5. Photosynthetic eﬃciency

The maximum photosynthetic rate (Amax), Vcmax,Jsat and TPU were significantly higher in treatments with K fertilization (Table 2). Throughfall exclusion did not significantly affect any parameters de-

rived from the A-Ci curves (P > 0.23). Mesophyll conductance (gm) was similar in all treatments except in -K + W, where it was signiﬁcantly lower (Table 2).

4. Discussion

4.1. Occurrence of hydraulic redistribution (HR)

In line with our first hypothesis, HR was observed in the studied Eucalyptus grandis plantation. The most unexpected result of this study was the presence of HR over the entire year, even during the rainy ff Fig. 4. E ect of water supply and potassium treatments on the seasonal var- seasons, when superficial soil layers were wet (Fig. 6). Just a few stu- iation in (a) leaf hydraulic conductance (K ) and (b) leaf-specific whole-tree leaf dies detected HR after large rain events (Burgess, 2006; Bleby et al., hydraulic conductance (K ), and (c) sapwood-specific K of Eucalyptus tree tree 2010), and hypothesized that it was the consequence of soil moisture grandis from October 2014 to May 2016 (56 to 72 months after planting). Applied treatments were: K fertilization and undisturbed rainfall (+K + W), no variability with some soil patches remaining dry after rainfall events. K fertilization and undisturbed rainfall (-K + W), no K application and 37% Interestingly, in some days we observed unexpected HR during the day, throughfall exclusion (-K-W), K fertilization and 37% throughfall exclusion which indicates that water potential gradients from shallow soil to leaf (+K-W). during canopy transpiration must be sometimes equivalent or slightly smaller than between soil layers within the rooting zone (Fig. S3). Since canopy transpiration was high at the same period that reverse flow was detected, daytime HR was not caused by cloudy conditions. Specifi- cally, we can hypothesize that daytime HR was probably localized in shallow roots that were highly connected to the deep taproot system. When water potentials gradients in the trunk and in the shallow roots reached equilibrium (probably before noon in Fig. S3), then water moved from deep soil layers to the trunk and then to the leaves, and in the meantime from deep soil roots to shallow roots. Daytime HR was observed in other tree species by Bleby et al. (2010), who suggested that it was caused by a mosaic of patchy dry soil, and the creation of such water potentials gradient but within shallow soil layers. Hultine et al. (2003) also suggested that daytime HR might occur in plants that are highly vulnerable to xylem embolism in order to prevent high loss Fig. 5. Average daily reverse sap flow (SF) velocity per month in horizontal of conductivity, which could be of importance for E. grandis since it has coarse roots of Eucalyptus grandis trees measured in the topsoil between been reported to be less tolerant to drought than sub-humid Eucalyptus November 2014 and April 2016 in the control (+W) and 37% throughfall ex- species (Bourne et al., 2017). Another hypothesis explaining HR during clusion (-W) plots, and with (+K) or without (-K) potassium addition. Values the day when tree transpiration rates were high could be linked to the are the mean ± SE (n = 180 or 186). Monthly values with different letters are significantly different (P < 0.05). existence of direct (physical) or indirect (mycorrhizae) mechanical connections between roots from the same tree or, but less likely (Laclau et al. personal communication), from two neighboring trees (Lev- potassium or water regime (Table 1). On the other hand, throughfall Yadun, 2011; Paula et al., 2015). If those anastomoses involved roots exclusion seemed to induce a decrease in the number of roots with RF with different absorption rates, then an imbalance flow could have been per month (Table S4), since more roots exhibited RF in the + W plots created between them, thus inducing reverse flow (Warren et al., 2008). than in -W plots, and so for almost all months except two from the wet Most of the HR measured in this study was probably water

7 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085

Fig. 6. Soil water storage from August 2014 (month 50) to April 2016 (month 70) in four diﬀerent soil layers ranging from the soil surface to a depth of 17 m (a, b, c and d) in the control (+W) and throughfall exclusion (-W) plots, with (+K) or without (-K) potassium addition. Water table depth over the same period in the + K+ W and + K-W plots is shown (e). Values are the mean ± SE (n = 3).

8 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085

Table 2

Eﬀects of potassium (+K) and 37 % throughfall exclusion (-W) on photosynthetic capacity parameters and mesophyll conductance (gm). Mean values were calculated from the A-Ci curves and normalized at 25 °C.

Treatment

-K + W +K + W -K-W +K-W

−2 −1 Amax (μmol CO2 m s ) 20.1 ± 0.9 b 28.2 ± 2.2 a 19.6 ± 1.7 b 29.3 ± 2.5 a −2 −1 Vc,max (μmol CO2 m s ) 82.1 ± 8.7 ab 100 ± 4.2 a 75.7 ± 7.8 b 103 ± 7.3 a −2 −1 Jsat (μmol CO2 m s ) 102 ± 6.1 b 138 ± 11 a 100 ± 8.6 b 147 ± 12 a −2 −1 TPU (μmol CO2 m s ) 5.3 ± 0.4 bc 7.6 ± 0.8 ab 4.8 ± 0.5 c 8.1 ± 0.9 a −2 −1 Rd (μmol CO2 m s ) 1.7 ± 0.01 a 1.7 ± 0.003 a 1.7 ± 0.003 a 1.7 ± 0 a −2 −1 −1 gm (μmol CO2 m s Pa ) 16.7 ± 7.9 b 30.1 ± 0.1 a 30.1 ± 0.1 a 30.0 ± 0 a

Different letters within a row indicate significantly differences between treatments (P < 0.05).

Amax: maximum photosynthetic rate at saturating PPFD; Vc,max: photosynthetic RuBisCO capacity per unit leaf area; Jsat: potential electron transport rate per unit leaf area; TPU: triose phosphate utilization per unit leaf area; Rd: rate of mitochondrial respiration in the light. hydraulically lifted (hydraulic lift) because top soil layers were drier transport capacity of K fertilized trees. than deep soil layers (Fig. 6), which likely created soil water potential K fertilization promoted an increase in wood production, which was gradients strong enough to move water up. In addition, for a given soil proportionally much higher than the increase in transpiration, thus layers, soil structure and root density were homogeneous and therefore strongly enhancing water-use efficiency (Fig. 2b). In fact, our WUEp water potential gradients at a given depth were probably small. Con- values in plots with + K, even with 37% throughfall exclusion, were sequently, the reverse flow that was measured in the topsoil (Fig. 5) was similar to the values observed in other Eucalyptus plantations without probably the consequence of water moving from deep to shallow roots water deficit (Forrester et al., 2010; Maier et al., 2017). This result and then to the shallow soil. indicates not only that potassium increased WUEp, but also that a de- The results did not agree with our second hypothesis: RF velocity crease of throughfall until 37 % did not significantly affect that effi- did not significantly increase in plots with potassium fertilization (+K) ciency. (Fig. 5). Another unexpected result was that trees under throughfall The annual tree transpiration did not increase over the period of six exclusion (−W) did not exhibit higher reverse flow (RF) than trees years in any treatment (the first three years are published in Battie- without (+W), and RF was even lower in −W in some of the months Laclau et al., 2016), and WUEp slightly increased in the second half of (Fig. 5, Table S2). We hypothesized that throughfall exclusion would the rotation with respect to the first in plots with potassium fertiliza- increase the amount of water transported by HR because of a decrease tion. Therefore, after six growing seasons, we observed that: 1) K fer- in water storage in superficial soil layers (Fig. 6). Our results could be tilization enhanced WUEp over time, and that 2) throughfall exclusion attributed to the fact that in all treatments, roots reached a permanent had no effect on WUEp. We postulate that the increment in WUEp may source of water by tapping directly into the water table (Johnson et al., result from a preferential partitioning of biomass toward stemwood, at 2014; Germon et al., 2019), which could explained why RF in some the expense of roots or leaves. At the same study site in Brazil, Epron roots was also sometimes quantified during the day (Fig S3). However, et al. (2012) showed that K fertilization increased both wood produc- if roots reached the water table, then why did hydraulic lift and re- tion and the fraction of carbon allocated aboveground in E. grandis distribution occur? On one hand, we hypothesize that the amount of plantations at the end of the rotation. According to the optimal parti- water they can take up from the water table is limited due to the low tioning theory (OPT), plants allocate additional biomass to the organ root biomass located at that depth (Germon et al., 2019). For that that takes up the resource that most limits growth (Bloom et al., 1985). reason, eucalypt trees still need to take up a considerable amount of Field studies showed that plants allocate relatively more carbon to water from other soil layers, as shown by Christina et al. (2017) at a shoots under light limitation and to roots under water and/or nutrient nearby site. In our study, HR probably helped keep the fine roots of limitation (Lapenis et al., 2013; Doughty et al., 2014; Girardin et al., eucalyptus alive during dry seasons, which is important to maintain a 2016). A change in the partitioning of biomass in favor of stemwood in high water and nutrients uptake capacity (Richards and Caldwell, 1987; the studied Eucalyptus stands fertilized with K was likely because of two Scholz et al., 2008), and to prevent embolism (Domec et al., 2004; Scott reasons. First, K fertilization increased leaf longevity (Battie-Laclau et al., 2008). The maintenance of a wide superficial root system also et al., 2013) and decreased limitations in the use of light through an allows a rapid uptake of soil water after small rainfall events (Burgess increase in the photosynthetic efficiency (Table 2), so trees with + K et al., 1998; Bleby et al., 2010). allocated less biomass in leaves at the end of the rotation. In fact, K Our results showed that neither potassium fertilization nor fertilization increased the ratio between wood increment to leaf area throughfall exclusion significantly affected the volumetric capacity of (Fig. 3c). Second, trees did not need to allocate too much biomass be- roots for transporting water away from the trunk (Table S2) nor the lowground because roots probably reached water table. Root biomass number of days with RF per month (Table S3). This reinforces the im- production in Eucalyptus growing on deep Ferralsols under tropical portance of long-term hydraulic redistribution monitoring, in addition climate is usually high in the first years of growth (Laclau et al., 2013; to root system trait measurements, for our understanding of water use Pinheiro et al., 2016), which probably contributed to decrease water in forests. supply limitation at the end of the rotation in our plantation.

4.2. Water use eﬃciency for biomass production (WUEp) 5. Conclusions

Despite a negative rainfall-transpiration balance was reached one Hydraulic redistribution (probably lift) occurred in the Eucalyptus month earlier in + K than in -K plots (data not shown), and soil water grandis plantation during the last two years of the rotation cycle, with or storage was lower in + K over the studied period (Fig. 6), tree tran- without potassium fertilization and 37% throughfall exclusion, and spiration was always higher under + K than in -K plots, likely as a even after large rain events. Neither potassium fertilization nor result of much higher LAI (Fig. S4a), which increases water demand, throughfall exclusion aﬀected the velocity of water transported by HR, and higher sapwood area (Fig. S4b), which increases the water probably because most trees reached water table. Potassium

9 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085 fertilization increased the water use efficiency for stemwood produc- Phytol. 203, 401–413. https://doi.org/10.1111/nph.12810. tion. This increase was associated to an increase in the photosynthetic Battie-Laclau, P., Laclau, J.-P., Piccolo M de, C., et al., 2013. Influence of potassium and ffi sodium nutrition on leaf area components in Eucalyptus grandis trees. Plant Soil 371, e ciency, LAI and sapwood area index, but not leaf nor whole-tree 19–35. https://doi.org/10.1007/s11104-013-1663-7. hydraulic conductance. Even if potassium fertilization can lead to Binkley, D., Campoe, O.C., Alvares, C., et al., 2017. The interactions of climate, spacing higher risk of hydraulic dysfunction in Eucalyptus under extreme water and genetics on clonal Eucalyptus plantations across Brazil and Uruguay. For. Ecol. fi Manage. 405, 271–283. https://doi.org/10.1016/j.foreco.2017.09.050. de cit, hydraulic redistribution can help fertilized trees to grow during Bleby, T.M., McElrone, A.J., Jackson, R.B., 2010. Water uptake and hydraulic redis- dry periods. Our results indicate that fertilizing E. grandis with po- tribution across large woody root systems to 20 m depth. Plant Cell Environ. 33, tassium is beneficial to increase both wood biomass production and 2132–2148. https://doi.org/10.1111/j.1365-3040.2010.02212.x. WUE , even with a 37% decrease in throughfall, when the deepest roots Bloom, A.J., Chapin, F.S., Mooney, H.A., 1985. Resource limitation in plants - an eco- p nomic analogy. Annu. Rev. Ecol. Syst. 16, 363–392. have access to water stored in the subsoil. Bourne, A.E., Creek, D., Peters, J.M.R., et al., 2017. Species climate range influences hydraulic and stomatal traits in Eucalyptus species. Ann. Bot. 123–133. https://doi. Authors' contributions org/10.1093/aob/mcx020. Brooks, J.R., Meinzer, F.C., Coulombe, R., Gregg, J., 2002. Hydraulic redistribution of soil water during summer drought in two contrasting Pacific Northwest coniferous for- JPL, JPB and YN conceived the experiment and provided the study ests. Tree Physiol. 22, 1107–1117. site. JPL, JPB, YN, CHAJ and VA got funding. VA and JDR implemented Brooksbank, K., Veneklaas, E.J., White, D.A., Carter, J.L., 2011. The fate of hydraulically redistributed water in a semi-arid zone eucalyptus species. Tree Physiol. 31, 649–658. the methodologies with the assistance of JPB, CHAJ and YN. VA and https://doi.org/10.1093/treephys/tpr052. JDR made data collection. VA made data curation and processing under Burgess, S., 2006. Redistribution of soil water by lateral roots mediated by stem tissues. J. the supervision of JCD, JDR and JG. VA made statistical analyses under Exp. Bot. 57, 3283–3291. https://doi.org/10.1093/jxb/erl085. Burgess, S.S.O., Adams, M.A., Turner, N.C., Ong, C.K., 1998. The redistribution of soil the supervision of JG. VA wrote the original draft of the manuscript and water by tree root systems. Oecologia 115, 306–311. https://doi.org/10.1007/ all authors review and edited it. All authors have read and approved the s004420050521. final version of the manuscript. Burgess, S.S.O., Adams, M.A., Turner, N.C., et al., 2001. Tree roots: conduits for deep recharge of soil water. Oecologia 126, 158–165. https://doi.org/10.1007/ s004420000501. Declaration of Competing Interest Cakmak, I., 2005. The role of potassium in alleviating detrimental effects of abiotic stresses in plants. J. Plant Nutr. Soil Sci. 168, 521–530. https://doi.org/10.1002/jpln. 200420485. The authors declare that they have no known competing financial Choat, B., Brodribb, T.J., Brodersen, C.R., et al., 2018. Triggers of tree mortality under interests or personal relationships that could have appeared to influ- drought. Nature 558, 531–539. https://doi.org/10.1038/s41586-018-0240-x. ence the work reported in this paper. Chou, S.C., Marengo, J.A., Lyra, A.A., et al., 2012. Downscaling of South America present climate driven by 4-member HadCM3 runs. Clim. Dyn. 38, 635–653. https://doi.org/ 10.1007/s00382-011-1002-8. Acknowledgements Christina, M., Le Maire, G., Battie-Laclau, P., et al., 2015. Measured and modeled inter- active effects of potassium deficiency and water deficit on gross primary productivity ffi – and light-use e ciency in Eucalyptus grandis plantations. Glob. Chang. Biol. 21, This work was funded by the São Paulo Research Foundation 2022–2039. https://doi.org/10.1111/gcb.12817. FAPESP (grant # 2013/25998-4, grant # 2018/13553-1), project USP/ Christina, M., Nouvellon, Y., Laclau, J.-P., et al., 2017. Importance of deep water uptake COFECUB 2011-25 / Uc Sv 134/12, and by the CIRAD. Last author in tropical eucalypt forest. Funct. Ecol. 31, 509–519. https://doi.org/10.1111/1365- fi 2435.12727. thanks the National Council for Scienti c and Technological Cochard, H., Bréda, N., Granier, A., 1996. Whole tree hydraulic conductance and water Development - CNPq for the Research Grant (grant # 312728/2017-4). loss regulation in Quercus during drought: evidence for stomatal control of embo- Authors specially thank to the students who helped with the leaf water lism? Ann des Sci For 53, 197–206. https://doi.org/10.1051/forest:19960203. ff David, T.S., Pinto, C.A., Nadezhdina, N., et al., 2013. Root functioning, tree water use and potential and biomass measurements. Authors thank to the sta at the hydraulic redistribution in Quercus suber trees: a modeling approach based on root Itatinga Experimental Station (ESALQ-USP), in particular Rildo Moreira sap flow. For. Ecol. Manage. 307, 136–146. https://doi.org/10.1016/j.foreco.2013. e Moreira (ESALQ-USP), as well as Eder Araújo da Silva and 07.012. FLORAGRO (www.floragroapoio.com.br) for their technical support. Delgado-Rojas, J.S., Laclau, J., Roupsard, O., et al., 2010. Calibration of home-made heat dissipation probes for a full rotation of Eucalyptus grandis trees in Brazil. AGU. p ID The experiment belongs to SOERE F-ORE-T, which is supported an- 972492. nually by Ecofor, Allenvi and the French National Research Delzon, S., Sartore, M., Granier, A., Loustau, D., 2004. Radial profiles of sap flow with – Infrastructure ANAEE-F (http://www.anaee-france.fr/fr/). We also ac- increasing tree size in maritime pine. Tree Physiol. 24, 1285 1293. https://doi.org/ 10.1093/treephys/24.11.1285. knowledge the support of the National Science Foundation, Award Djaman, K., Irmak, S., Martin, D.L., et al., 2013. Plant nutrient uptake and soil nutrient Number: NSF‐IOS‐1754893. The authors also thank to the editor and dynamics under full and limited irrigation and rainfed maize production. Agron. J. – the anonymous reviewer for their valuable comments, which helped to 105, 527 538. https://doi.org/10.2134/agronj2012.0269. Domec, J.-C., King, J.S., Noormets, A., et al., 2010. Hydraulic redistribution of soil water improve the quality of the manuscript. by roots affects whole-stand evapotranspiration and net ecosystem carbon exchange. New Phytol. 187, 171–183. https://doi.org/10.1111/j.1469-8137.2010.03245.x. Appendix A. Supplementary data Domec, J.-C., Noormets, A., King, J.S., et al., 2009. Decoupling the influence of leaf and root hydraulic conductances on stomatal conductance and its sensitivity to vapour pressure deficit as soil dries in a drained loblolly pine plantation. Plant Cell Environ. Supplementary material related to this article can be found, in the 32, 980–991. https://doi.org/10.1111/j.1365-3040.2009.01981.x. online version, at doi:https://doi.org/10.1016/j.envexpbot.2020. Domec, J.-C., Warren, J.M., Meinzer, F.C., et al., 2004. Native root xylem embolism and stomatal closure in stands of Douglas-fir and ponderosa pine: mitigation by hydraulic 104085. redistribution. Oecologia 141, 7–16. https://doi.org/10.1007/s00442-004-1621-4. Doughty, C.E., Malhi, Y., Araujo-Murakami, A., et al., 2014. Allocation trade-offs dom- References inate the response of tropical forest growth to seasonal and interannual drought. Ecology 95, 2192–2201. https://doi.org/10.1890/13-1507.1. Epron, D., Laclau, J.-P., Almeida, J.C.R., et al., 2012. Do changes in carbon allocation Almeida, J.C.R., Laclau, J.-P., Gonçalves, J.L., de, M., et al., 2010. A positive growth account for the growth response to potassium and sodium applications in tropical response to NaCl applications in Eucalyptus plantations established on K-deficient Eucalyptus plantations? Tree Physiol. 32, 667–679. https://doi.org/10.1093/ soils. For. Ecol. Manage. 259, 1786–1795. https://doi.org/10.1016/j.foreco.2009.08. treephys/tpr107. 032. Farquhar, G., Hubick, K., Condon, A., Richards, R., 1989. Carbon isotope fractionation Arquero, O., Barranco, D., Benlloch, M., 2006. Potassium starvation increases stomatal and plant water-use efficiency. In: Rundel, P., Ehleringer, J., Nagy, K. (Eds.), Stable conductance in olive trees. HortScience 41, 433–436. Isotopes in Ecological Research. Springer, Heidelberg, Germany & New York, USA, Battie-Laclau, P., Delgado-Rojas, J.S., Christina, M., et al., 2016. Potassium fertilization pp. 21–46. increases water-use efficiency for stem biomass production without affecting intrinsic Farquhar, G.D., von Caemmerer, S., Berry, J.A., 1980. A biochemical model of photo- water-use efficiency in Eucalyptus grandis plantations. For. Ecol. Manage. 364, synthetic CO2 assimilation in leaves of C3 species. Planta 149, 78–90. https://doi. 77–89. https://doi.org/10.1016/j.foreco.2016.01.004. org/10.1007/BF00386231. Battie-Laclau, P., Laclau, J.-P., Domec, J.-C., et al., 2014. Effects of potassium and sodium Forrester, D.I., Collopy, J.J., Morris, J.D., 2010. Transpiration along an age series of supply on drought-adaptive mechanisms in Eucalyptus grandis plantations. New Eucalyptus globulus plantations in southeastern Australia. For. Ecol. Manage. 259,

10 V. Asensio, et al. Environmental and Experimental Botany 176 (2020) 104085

1754–1760. https://doi.org/10.1016/j.foreco.2009.04.023. isotopic composition, leaf gas exchanges and plant growth. Plant Physiol. Biochem. Germon, A., Jourdan, C., Bordron, B., et al., 2019. Consequences of clear-cutting and 137, 102–112. https://doi.org/10.1016/j.plaphy.2019.02.006. drought on fine root dynamics down to 17 m in coppice-managed eucalypt planta- Melcher, P.J., Meinzer, F.C., Yount, D.E., et al., 1998. Comparative measurements of tions. For. Ecol. Manage. 445, 48–59. https://doi.org/10.1016/j.foreco.2019.05.010. xylem pressure in transpiring and non-transpiring leaves by means of the pressure Gessler, A., Schaub, M., McDowell, N.G., 2017. The role of nutrients in drought-induced chamber and the xylem pressure probe. J. Exp. Bot. 49, 1757–1760. https://doi.org/ tree mortality and recovery. New Phytol. 214, 513–520. https://doi.org/10.1111/ 10.1093/jxb/49.327.1757. nph.14340. Moura, M.V.P., da, S., Farias, C.H., de, A., de Azevedo, C.A.V., et al., 2005. Doses de Getirana, A., 2016. Extreme water deficit in Brazil detected from space. J. Hydrometeorol. adubação nitrogenada e potássica em cobertura na cultura da cana-de-açúcar, pri- 17, 591–599. https://doi.org/10.1175/JHM-D-15-0096.1. meira soca, com e sem irrigação. Ciência e Agrotecnologia 29, 753–760. https://doi. Girardin, C.A.J., Malhi, Y., Doughty, C.E., et al., 2016. Seasonal trends of Amazonian org/10.1590/S1413-70542005000400006. rainforest phenology, net primary productivity, and carbon allocation. Global Nadezhdina, N., Ferreira, M.I., Conceição, N., et al., 2015. Water uptake and hydraulic Biogeochem. Cycles 30, 700–715. https://doi.org/10.1002/2015GB005270. redistribution under a seasonal climate: long-term study in a rainfed olive orchard. Gonçalves, J., Stape, J., Laclau, J.-P., et al., 2008. Assessing the effects of early silvi- Ecohydrology 8, 387–397. https://doi.org/10.1002/eco.1545. cultural management on long-term site productivity of fast-growing eucalypt plan- Noormets, A., Nouvellon, Y., 2015. Introduction for special issue: carbon, water and tations: the Brazilian experience. South For. J. For. Sci. 70, 105–118. https://doi.org/ nutrient cycling in managed forests. For. Ecol. Manage. 355, 1–3. https://doi.org/10. 10.2989/SOUTH.FOR.2008.70.2.6.534. 1016/j.foreco.2015.08.022. Gonçalves, J.L.D.M., Alvares, C.A., Higa, A.R., et al., 2013. Integrating genetic and sil- Paula, R.R., Bouillet, J.-P., Ocheuze Trivelin, P.C., et al., 2015. Evidence of short-term vicultural strategies to minimize abiotic and biotic constraints in Brazilian eucalypt belowground transfer of nitrogen from Acacia mangium to Eucalyptus grandis trees plantations. For. Ecol. Manage. 301, 6–27. https://doi.org/10.1016/j.foreco.2012. in a tropical planted forest. Soil Biol. Biochem. 91, 99–108. https://doi.org/10.1016/ 12.030. j.soilbio.2015.08.017. Granier, A., 1985. Une nouvelle méthode pour la mesure du flux de séve brute dans le Pinheiro, R.C., de Deus, J.C., Nouvellon, Y., et al., 2016. A fast exploration of very deep tronc des arbres. Ann des Sci For 42, 81–88. soil layers by Eucalyptus seedlings and clones in Brazil. For. Ecol. Manage. 366, Harley, P.C., Thomas, R.B., Reynolds, J.F., Strain, B.R., 1992. Modelling photosynthesis of 143–152. https://doi.org/10.1016/j.foreco.2016.02.012. cotton grown in elevated CO2. Plant Cell Environ. 15, 271–282. https://doi.org/10. Pradier, C., Hinsinger, P., Laclau, J.-P., et al., 2017. Rainfall reduction impacts rhizo- 1111/j.1365-3040.1992.tb00974.x. sphere biogeochemistry in eucalyptus grown in a deep Ferralsol in Brazil. Plant Soil Harvey, H.P., van den Driessche, R., 1997. Nutrition, xylem cavitation and drought re- 414, 339–354. https://doi.org/10.1007/s11104-016-3107-7. sistance in hybrid poplar. Tree Physiol. 17, 647–654. https://doi.org/10.1093/ Richards, J.H., Caldwell, M.M., 1987. Hydraulic lift: substantial nocturnal water transport treephys/17.10.647. between soil layers by Artemisia tridentata roots. Oecologia 73, 486–489. https:// Hultine, K.R., Cable, W.L., Burgess SSO, Williams D.G., 2003. Hydraulic redistribution by doi.org/10.1007/BF00379405. deep roots of a Chihuahuan Desert phreatophyte. Tree Physiol. 23, 353–360. https:// Robin, A., Pradier, C., Sanguin, H., et al., 2019. How deep can ectomycorrhizas go? A case doi.org/10.1093/treephys/23.5.353. study on Pisolithus down to 4 meters in a Brazilian eucalypt plantation. Mycorrhiza IPCC, 2014. Climate Change 2014: Synthesis Report. Geneva, Switzerland. . 29, 637–648. https://doi.org/10.1007/s00572-019-00917-y. Johnson, D.M., Brodersen, C.R., Reed, M., et al., 2014. Contrasting hydraulic architecture Salazar, A., Katzfey, J., Thatcher, M., et al., 2016. Deforestation changes land–atmosphere and function in deep and shallow roots of tree species from a semi-arid habitat. Ann. interactions across South American biomes. Glob. Planet. Change 139, 97–108. Bot. 113, 617–627. https://doi.org/10.1093/aob/mct294. https://doi.org/10.1016/j.gloplacha.2016.01.004. Kunert, N., Schwendenmann, L., Potvin, C., Hölscher, D., 2012. Tree diversity enhances Scholz, F.G., Bucci, S.J., Goldstein, G., et al., 2008. Biophysical and life-history de- tree transpiration in a Panamanian forest plantation. J. Appl. Ecol. 49, 135–144. terminants of hydraulic lift in Neotropical savanna trees. Funct. Ecol. 22, 773–786. https://doi.org/10.1111/j.1365-2664.2011.02065.x. https://doi.org/10.1111/j.1365-2435.2008.01452.x. Laclau, J.-P., Almeida, J.C.R., Gonçalves, J.L.M., et al., 2009. Influence of nitrogen and Scholz, F.G., Bucci, S.J., Goldstein, G., et al., 2002. Hydraulic redistribution of soil water potassium fertilization on leaf lifespan and allocation of above-ground growth in by neotropical savanna trees. Tree Physiol. 22, 603–612. Eucalyptus plantations. Tree Physiol. 29, 111–124. https://doi.org/10.1093/ Scolforo, H.F., McTague, J.P., Burkhart, H., et al., 2019a. Yield pattern of eucalypt clones treephys/tpn010. across tropical Brazil: an approach to clonal grouping. For. Ecol. Manage. 432, 30–39. Laclau, J.-P., Ranger, J., de Moraes Gonçalves, J.L., et al., 2010. Biogeochemical cycles of https://doi.org/10.1016/j.foreco.2018.08.051. nutrients in tropical Eucalyptus plantations. Main features shown by intensive Scolforo, H.F., McTague, J.P., Burkhart, H., et al., 2019b. Modeling whole-stand survival monitoring in Congo and Brazil. For. Ecol. Manage. 259, 1771–1785. https://doi.org/ in clonal eucalypt stands in Brazil as a function of water availability. For. Ecol. 10.1016/j.foreco.2009.06.010. Manage. 432, 1002–1012. https://doi.org/10.1016/j.foreco.2018.10.044. Lapenis, A.G., Lawrence, G.B., Heim, A., et al., 2013. Climate warming shifts carbon al- Scott, R.L., Cable, W.L., Hultine, K.R., 2008. The ecohydrologic significance of hydraulic location from stemwood to roots in calcium-depleted spruce forests. Glob. redistribution in a semiarid savanna. Water Resour. Res. 44, W02440. https://doi. Biogeochem. Cycles 27, 101–107. https://doi.org/10.1029/2011GB004268. org/10.1029/2007WR006149. Law, B., Falge, E., Gu, L., et al., 2002. Environmental controls over carbon dioxide and Sharkey, T.D., Bernacchi, C.J., Farquhar, G.D., Singsaas, E.L., 2007. Fitting photo- water vapor exchange of terrestrial vegetation. Agric. For. Meteorol. 113, 97–120. synthetic carbon dioxide response curves for C3 leaves. Plant Cell Environ. 30, https://doi.org/10.1016/S0168-1923(02)00104-1. 1035–1040. https://doi.org/10.1111/j.1365-3040.2007.01710.x. Lev-Yadun, S., 2011. Why should trees have natural root grafts? Tree Physiol. 31, Sherwood, S., Fu, Q., 2014. A drier future? Science 80 (343), 737–739. https://doi.org/ 575–578. https://doi.org/10.1093/treephys/tpr061. 10.1126/science.1247620. Lobell, D.B., Schlenker, W., Costa-Roberts, J., 2011. Climate trends and global crop Stape, J., Binkley, D., Ryan, M.G., 2004. Eucalyptus production and the supply, use and production since 1980. Science 80 (333), 616–620. https://doi.org/10.1126/science. efficiency of use of water, light and nitrogen across a geographic gradient in Brazil. 1204531. For. Ecol. Manage. 193, 17–31. https://doi.org/10.1016/j.foreco.2004.01.020. Long, S.P., Bernacchi, C.J., 2003. Gas exchange measurements, what can they tell us Stape, J.L., Binkley, D., Ryan, M.G., et al., 2010. The Brazil Eucalyptus Potential about the underlying limitations to photosynthesis? Procedures and sources of error. Productivity Project: influence of water, nutrients and stand uniformity on wood J. Exp. Bot. 54, 2393–2401. https://doi.org/10.1093/jxb/erg262. production. For. Ecol. Manage. 259, 1684–1694. https://doi.org/10.1016/j.foreco. Mahmood, T., Saeed, M., Ahmad, R., Ghaffar, A., 1999. Water and potassium manage- 2010.01.012. ment for enhanced maize (Zea mays L.) productivity. Int. J. Agric. Biol. 1, 314–317. Warren, J.M., Brooks, J.R., Meinzer, F.C., Eberhart, J.L., 2008. Hydraulic redistribution of Maier, C.A., Albaugh, T.J., Cook, R.I., et al., 2017. Comparative water use in short-ro- water from Pinus ponderosa trees to seedlings: evidence for an ectomycorrhizal tation Eucalyptus benthamii and Pinus taeda trees in the Southern United States. For. pathway. New Phytol. 178, 382–394. https://doi.org/10.1111/j.1469-8137.2008. Ecol. Manage. 397, 126–138. https://doi.org/10.1016/j.foreco.2017.04.038. 02377.x. Mateus, N., de, S., Ferreira EV de, O., Arthur Junior, J.C., et al., 2019. The ideal per- Weaver, J.E., 1919. The Ecological Relations of Roots. Carnegiei Institution of centage of K substitution by Na in Eucalyptus seedlings: evidences from leaf carbon Washington, Washington D.C.